Role of Performance Evaluator in data-driven Fault Tolerant Control

Abstract

In this note, we discuss about the performance evaluator tool that plays a significant role in the data-driven fault tolerant control (FTC) system. In our notion of data-driven approach, we do not have any access to the a priori plant model in real-time. Moreover, we do not equip any estimation algorithm to determine the model of the plant. Here, we use the trajectories generated by the system in real time. These trajectories, in fact, capture the behavior of the system and we directly evaluate the control performance of the closed-loop based on these trajectories. Whenever a fault occurs, this tool assists the supervisor to take necessary actions for the controller reconfiguration mechanism. Since no a priori knowledge about the plant is used online, this tool is shown here performing all the necessary roles, particularly fault detection, fault accommodation, and especially, stability assessment.